9633

• UCREFRP Catalog Number: 9633
• Author: Schmidt, J. C. and J. B. Box
• Title: Application of a Dynamic Model to Assess Controls on Age-0 Colorado Pikeminnow Distribution in the Middle Green River, Colorado and Utah
• USFW Year: 2004
• USFW - Doc Type: Annals of the Association of American Geographers
• Copyright Material: NO

Colorado Pikeminnow Distribution

467

shoreline complexity is a poor predictor of backwater
area, because the metric is sensitive to the number of
islands in a reach, and islands do not necessarily create
backwaters. Our use of (2) avoids the difficulty of pre-
dicting backwater area from Sc.
Unfortunately, the shape of the actual relationship
between the proportion of the main flow lined by sepa-
ration surfaces and SC is unknown and may differ from
reach to reach and with temporal changes in bar top-
ography. We assumed a linear relationship:

b = 0.15 SC - 0.3.

(3)

The intercept of this relation was defined such that b = 0
when SC = 2 (the minimum possible value of SC). The
slope of the relation was arbitrarily chosen, because
relevant field measurements are not available. The value
of b varies between 0 and 0.6 for typical values of Sc.
We based these assumptions on field studies of the
longitudinal dispersion of dye because the behavior of
neutrally buoyant particles, and thus drifting larvae, is
similar to that of dye clouds (Fischer 1973). The ex-
change of water with backwaters, called "dead zones," is
responsible for the time lag between passage of the peak
concentration, centroid of mass, and trailing edge of
longitudinally-dispersing dye clouds in natural rivers
(Fischer et al. 1979). The cumulative effect of the ex-
change of water with backwaters is partly determined by
geomorphic attributes of the channel such as SC, be-
cause skewness of the dye cloud increases as the size and
number of dead zones increases (Valentine and Wood
1977). For example, reaches with more complex shore-
line topography and more backwaters had slower rates of
downstream drift of neutrally buoyant particles on the
Pecos River in New Mexico (Dudley and Platania
2000b). In debris fan-affected canyons, parts of recir-
culating eddies become dead zones at low flow when
eddy bars are emergent. However, recirculating eddies in
the Colorado River in Grand Canyon are not dead zones,
and dye clouds display less longitudinal dispersion than
in alluvial rivers (Graf 1995).
Estimates ofb were made for the 10 reaches, based on
measurements of SC (Figure 7). These measurements of
SC were originally made to characterize habitat com-
plexity in 575 km of the Green River between the
Colorado River confluence and the upstream end of
reach D (Schmidt 1994) and were made from aerial
photographs (nominal scale 1:2000) taken in late sum-
mer 1963 at base flows between 10 and 28 m3 s -1 in
alternating 15-km reaches. SC measurements for four of
these reaches, comprising 35 percent of the study area
(Table 3), were taken from these data. Elsewhere, we
used Schmidt's (1994) Green River average SC values

for three planform types. These planform types were:
meandering with a wide alluvial valley (called restricted
meanders by Ikeda [1989]), meandering with a narrow
alluvial valley (called fixed meanders by Ikeda [1989]),
or debris fan-affected canyons where fan-eddy complexes
are numerous (Schmidt and Rubin 1995). We assumed
no change in river geomorphology since 1963, and we
assumed that the relationship was the same in all years.
We made assumptions about how SC changes at
discharges higher than base flow using geomorphic sur-
rogates: the shoreline defined by the contact of wet and
dry sand and the shoreline at bankfull stage (Figure 7).
Rakowski (1997) found that the former attribute is a
consistent feature in various years of photography of the
study area, and many studies of sand bar characteristics
in the Colorado River in Grand Canyon have used this
surrogate (Schmidt, Grams, and Leschin 1999; Goeking,
Schmidt, and Webb 2003).
The shape of the relationship between discharge and
SC was assumed to be unimodal (Figure 7C) and was
estimated in the following way. We assumed that com-
plexity is a minimum value of 2 when discharge is O. We
assumed that the discharge associated with the contact
between wet and dry sand was 33 percent of the differ-
ence between base flow and bankfull discharge. This
assumption may overpredict backwater habitat at the
discharges of concern to this model because Rakowski
(1997) determined that the contact between wet and
dry sand varied between 19 and 22 percent for aerial
photographs taken in the early 1990s. We assumed that
SC was the same between 0.8 and 1.0 times the bankfull
discharge, consistent with the findings of Rakowski
(1997). The magnitude of bankfull stage was estimated
to be the discharge of the post-dam, two-year recurrence
flood. Over-bank conditions are not relevant to the
modeling exercise.

Biology. As described in (2), we assumed that larvae
intentionally swim into backwaters as well as being
transported there by the current, and we parameterized
this process as a function of the proportion of the pop-
ulation with the ability to swim. We assumed that the
ability to swim was directly proportional to the growth
rate of the fish because Paulin, Williams, and Tyus
(1989) found that swimming ability is linearly related
to fish size. Pikeminnow less than 10 mm in length can-
not swim upstream against velocities of 10 cm sec - 1,
and at 12.2 mm length they begin to control their po-
sition in river channels (Paulin et al. 1988). The rate of
growth of pike minnow was assumed to be

1 = 0.1l4d + 8.886

(4)